An approach to expert systems for image processing software libraries

Large scientific software libraries are used in various application areas in order to reduce the scientists' programming work. However, the application of software packages is not always easy and often requires profound knowledge about the application domain and the package. Expert systems can be used to support less experienced users of software libraries. This paper presents a general approach to the construction of such expert systems and describes two prototype systems for the application of the SPIDER and GIPSY image processing packages. The central idea of our approach lies in the subdivision of the knowledge base into three parts containing knowledge about the use of software packages in general, image processing, and the actually applied software package, respectively.